lars-quaedvlieg / optml_course

EPFL Course - Optimization for Machine Learning - CS-439

EPFL Course - Optimization for Machine Learning - CS-439

Official coursebook information

Lectures: Fri 13:15-15:00 in CO2

Exercises: Fri 15:15-17:00 in BC01

This course teaches an overview of modern mathematical optimization methods, for applications in machine learning and data science. In particular, scalability of algorithms to large datasets will be discussed in theory and in implementation.

Team

• Instructors:
  • Martin Jaggi [email protected]
  • Nicolas Flammarion [email protected]
• Assistants:
  • Aditya Varre [email protected]
  • Atli Kosson [email protected]
  • Dongyang Fan [email protected]
  • Oguz Kaan Yüksel [email protected]

Contents:

Convexity, Gradient Methods, Proximal algorithms, Subgradient Methods, Stochastic and Online Variants of mentioned methods, Coordinate Descent, Frank-Wolfe, Accelerated Methods, Primal-Dual context and certificates, Lagrange and Fenchel Duality, Second-Order Methods including Quasi-Newton Methods, Derivative-Free Optimization.

Advanced Contents:

Parallel and Distributed Optimization Algorithms, Federated Learning

Computational Trade-Offs (Time vs Data vs Accuracy), Lower Bounds

Non-Convex Optimization: Convergence to Critical Points, Alternating minimization, Neural network training

Program:

Nr	Date	Topic	Materials	Exercises
1	23.2.	Introduction, Convexity	notes, slides	lab00
2	1.3.	Gradient Descent	notes, slides	lab01
3	8.3.	Projected Gradient Descent	notes, slides	lab02
4	15.3.	Proximal and Subgradient Descent	notes, slides	lab03
5	22.3.	Stochastic Gradient Descent, Non-Convex Optimization	notes, slides	lab04
.	29.3.	easter vacation		-
.	5.4.	easter vacation		-
6	12.4.	Non-Convex Optimization	notes, slides	lab05
7	19.4.	Newton's Method & Quasi-Newton	notes, slides	lab06
8	26.4.	Coordinate Descent	notes, slides	lab07
9	3.5.	Frank-Wolfe	notes, slides	lab08
10	10.5.	Mini-Project week		-
11	17.5.	Accelerated Gradient, Gradient-free, adaptive methods	notes, slides	lab09
12	24.5.	Opt for ML in Practice	notes, slides	lab10
13	31.5.	Opt for ML in Practice	notes, slides	Q&A Projects

Videos:

• Playlist of 2024 videos (EPFL internal)
• Public playlist of 2021 videos (youtube)

Exercises:

The weekly exercises consist of a mix of theoretical and practical Python exercises for the corresponding topic each week (starting week 2). Solutions to exercises are available in the lab folder.

Discussion forum (EPFL internal)

Project:

A mini-project will focus on the practical implementation: Here we encourage students to investigate the real-world performance of one of the studied optimization algorithms or variants, helping to provide solid empirical evidence for some behaviour aspects on a real machine-learning task. The project is mandatory and done in groups of 3 students. It will count 30% to the final grade. Project reports (3 page PDF) are due June 14th. Here is a detailed project description.

Assessment:

Session Exam on Monday 01.07.2024 from 15h15 to 18h15 (INM200, INM202). Format: Closed book. Theoretical questions similar to exercises. You are allowed to bring one cheat sheet (A4 size paper, both sides can be used).

For practice:

• exams 2023, 2022, 2021, 2020, 2019, 2018
• solutions 2023, 2022, 2021, 2020, 2019, 2018.

Links to related courses and materials

• CMU 10-725
• Berkeley EE-227C

Recommended Books

• Convex Optimization: Algorithms and Complexity, by Sébastien Bubeck (free online)
• Convex Optimization, Stephen Boyd and Lieven Vandenberghe (free online)
• Introductory Lectures on Convex Optimization, Yurii Nesterov (free online)